This community-based, cross-sectional study will test a simple low-cost diagnostic modality for rapid diagnosis of pulmonary tuberculosis (TB). We shall test whether saline-induced sputum improves the diagnostic yield of smear positive TB in HIV-infected and HIV-uninfected adolescents and adults. It will be tested at our established research site near Cape Town, South Africa, where clinical and epidemiological studies are being conducted in preparation for trials of novel TB vaccines. We have successfully used this technique to investigate over 1,800 children in a Phase IV BCG vaccine trial. Specific aims: To determine whether induced sputum improves rapid microbiological diagnosis of pulmonary TB in: (a) HIV-infected people, and (b), in all people with suspected tuberculosis, regardless of HIV status. We shall enroll 600 patients presenting to community health services with suspected TB. Two routine sputum collections, one induced sputum collection with saline nebulization, and an HIV test will be performed. Direct smear sputum microscopy and mycobacterial culture will be performed in an accredited laboratory. HIV test results will be used to stratify participants for evaluation of the primary outcome, positive direct smear microscopy, and the secondary outcome, positive culture of Mycobacterium tuberculosis. These outcomes will be analyzed as the difference in diagnostic yield (proportion of culture-proven TB cases) detected by direct smear microscopy of one, or two, routine sputum collections, compared to a single induced sputum. TB accounted for 10% of all deaths in this region in 2004. Rapid diagnosis of pulmonary TB is especially critical in HIV-infected individuals, since delay in therapy may be fatal. Paradoxically, diagnosis is more difficult in HIV-infected people with pauci-bacillary disease. Previous studies have demonstrated diagnostic benefit of induced sputum among individuals with negative routine sputum smear microscopy. This study will allow objective assessment of the impact of induced sputum on TB case detection among HIV-infected and HIV-uninfected persons, by calculating the gain in diagnostic yield that might be achieved by a regional TB control program in a high-prevalence developing country.